Okay, I’m going to start a new topic revising (my own) old post about blending paints. I’ve found myself oddly drawn to solving this question with an objective method, that is, trying to match a given color by blending specific colors of a specific paint using some sort of numeric data. I have several different brands of paint that I use - not all equally - and occasionally I’m looking for a small quantity of a color and either:
can’t find it made by a manufacturer (that I use), or
don’t want to buy a large bottle/jar when I need 2mL (nobody needs 30mL of FS 32246 Navy Torpedo).
I was using iModelKit (iOS app, not sure if it’s available via Google store) to do some calculations, but started pulling my own information from various sources to make a spreadsheet for calculations. The big bugaboo is that (as y’all know), mixing LIGHT colors is different from mixing PIGMENT colors, and some paints use pigments that mix differently than the same color paint that uses a different pigment…why sometimes red + blue = brown, instead of purple.
Anyway, all this said, has anyone else done anything like this, so maybe I can pull someone else into the discussion to bounce thoughts off of?
Some of those issues sound like problems with how the manufacturer makes their colors. If you’re getting brown out of red and blue it sounds like there is some yellow in one of them. Most likely in the red.
I do mix a lot of my own colors. I do enjoy the ease of just grabbing a color, but most of the time what I want isn’t a pre-made color.
I found studying color theory to be the best way to figure out how to achieve the color I am looking for.
So the goal for me has been to determine three things:
How to characterize known paints and their colors numerically
How to characterize a desired color numerically
How to calculate blending of a multiple of (1) to achieve (2)
The first isn’t ridiculously hard to do, although the initial color characterization by hex code or LAB led me to realize that to do so properly, you really need a color spectra, which is more data than just a single code. I’ve found some sources of this data, although it looks like the iModelKit app has (potentially) proprietary data for this specific to the actual model paints, versus the more readily available artist oil paints/etc.
The second item follows the first, except it’s harder to find data for a very specific color. If I keep running headlong down this rabbit trail without any caution, I may wind up purchasing a handheld spectrometer so I can test samples of known colors, but most of those would have to be chips for what I’m looking for. Luckily, I have quite a few chips and color fan references.
Finally, the third is where things get really interesting. I’ve already gone deeeep down a rabbit hole on color calculation by weighted geometric mean of reflectance curves, Kubelka-Munk theory, and the existence of several different software options for mixing calculations, which of course I’m looking to reverse-engineer to determine how they work!
In what might be a fortunate shift in circumstances, I think I’ll actually have a bit more time now to actually…you know…build models…but I’m going to keep thinking about this. I’ve purchased a book or two about mixing of traditional artists mediums (oil paints, etc) that help a little with where different pigments shift in the color spectra when mixed with others, but I think I may very seriously consider making my own color wheel.
Nothing like just brute-forcing your way to something usable.
Pantone Colors are the industry standard, at least in graphic design
Pantone can be converted to CMYK (Cyan, Magenta, Yellow, Black) blend (usually a percentage of each) and you can mix your paints with those colors. Same is true for RGB (Red Green Blue)
There are calculators out there for the conversion of Pantone to CMYK. This is a quick one I found.
There are also RGB conversions out there as well. Honestly, if you can find extremely good Red, Green, Blue paints or Cyan, Magenta, Yellow, and Black you can mix any pretty much any classified color out there.
If Pantone isn’t broad enough then go fully digital and use the colorimeter as your color picker. Load into just about any relatively decent modern digital Photo Image editor and they will convert it to a CMYK or RGB blend for you.
Thanks, @PhoenixG, that’s spot on with where my path has taken me. In fact, I wound up purchasing (but haven’t really used much) a Pantone Color Match Card for ~$15 because I wasn’t interested in paying a fortune for a Pantone book (or books).
What I ran into when I started going down the RGB/CMYK path is that the calculations for a blend didn’t always work, usually because one or more of the paints seemed to be using a pigment that carried other color implications. The most egregious example was I once tried to mix up a green shade and got some sort of hideous blackish mess that…was not green!
In short, what I’ve found is that using a ‘simple’ sRGB or CMYK “code” didn’t bring with it enough information for the math to work all the time. The best example is that a true “white” has CMYK values of 0 0 0 0 because it relies on the underlying paper substrate! If I tried to calculate a shade from a particular hue, the math would give me the lighter shade, but I’d have nothing to blend to achieve that shade, other than perhaps a light airbrush application over a white primer. All of this is conflicting with the “pigment coatings are opaque” mindset I’ve developed.
From what I’ve read so far down one of these rabbit trails, the use of the handheld colorimeter to characterize a paint sample with a spectra across a wavelength range of ~380-740 provides enough data that (again, theoretically) the math to do a blend will produce a reasonably close result. As I have more time to get to this, I’m going to use publicly available data to prove to myself that I can achieve target colors with that math, but I think I’m going to build something first and then come back to this. I never intended to develop a hobby that takes me away from my hobby!
I’ll add to this that when I started, I was trying to find a target color by blending paints that I had on hand. What I’m finding is that I’ll likely be better off blending from a limited number of “primary” and “secondary” paint colors from a line and not trying to take (e.g.) RLM04 + RLM24 = RLM25!
The complexity of making colors is probably why mfg for so long produced limited lines. I don’t know if there are any consumer level products that will ever allow you to achieve the goal of perfect colors every time, but I wish you luck on your quest.
Thanks for the suggestion! I have a CMYK wheel, which I find is more applicable to pigments, but it’s only vaguely helpful - mostly for helping me overcome my “light mixing” (vs. “pigment mixing”) prejudice.
On an unrelated note, yesterday I discovered that I might actually need a pretty reasonable quantity of FS32246 Navy Torpedo and I cannot stop laughing at myself (please see OP)!
